[Cryptography] "Trust in digital certificate ecosystem eroding"

[Bill Frantz]

On 5/6/15 at 2:47 PM, cat at reptiles.org (Cat Okita) wrote:

>On Mon, 4 May 2015, Bill Frantz wrote:
>>We, as engineers, need to present security information to our 
>>users in a way that is meaningful to them. They might be more 
>>concerned about a revoked certificate than about an expired 
>>one, just as they might be about a driver's license. They 
>>might want to know the chain of trust they are depending on, 
>>but we don't tell them either of these things. If we show them 
>>the MITM certificates, they will be in a much better position 
>>to judge how much trust to place in the connection. If we show 
>>them the convoluted trust chain, the organizations depending 
>>on those chains may decide to make the users decision easier 
>>by cleaning up their acts. And enough users will look at this 
>>information in the same way they check businesses with the 
>>chamber of commerce and friends living in the community.
>
>Why should the user care about any of that?  What we -should- be presenting
>is something with the equivalent clarity of "Look, there are cars rushing
>by my face, -and- the light is red -- perhaps I shouldn't walk across the
>road", rather than "you have a blindfold on, and there are some noises
>that suggest that you might be able to cross the road in some direction,
>but we're going to presume that you already know what they mean, and what's
>the best way to actually proceed".

I think we are in violent agreement about what is needed. We 
need to speak the user's language.


>In order to make good (or even tolerable) decisions about acceptable
>risk, you have to have enough context to have the faintest clue what sort
>of risks you're actually looking at (see also:  "I have a system that
>totally works for the stock market!").

Note that the users are the only ones who have the context 
necessary to make these decisions. We need to show our users 
which institutions they are depending on for the security 
assertions. They can make the decision of whether these 
institutions are trustworthy enough for the current context. In 
some contexts, e.g. reading a blog post, not much assurance is 
needed. More is needed for online banking. While we probably 
don't need to show this information unless asked, we do need to 
make it available.


>I don't have any answers for the 'best' way to approach this, but providing
>more technical detail in the absence of well understood grounds 
>for evaluating said detail definitely isn't it.

Technical detail isn't needed. Only geeks like us care if the 
hash is MD5 or SHA512. Until we see actual attacks against one 
of them in the field, they are equivalent. As geeks, we know 
that MD5 has significant weaknesses, and we need to work to 
remove MD5 from use, but that is background work that doesn't 
need to be seen in the UI. It is like upgrading physical locks 
to use pins along with wards instead of just using wards.

Cheers - Bill

-------------------------------------------------------------------------
Bill Frantz        | The first thing you need when  | Periwinkle
(408)356-8506      | using a perimeter defense is a | 16345 
Englewood Ave
www.pwpconsult.com | perimeter.                     | Los Gatos, 
CA 95032